Unsupervised Learning

Mounic Madiraju

Published Oct 24, 2020

What is Unsupervised Learning?

Unsupervised learning is a machine learning technique that does not require users to supervise a model. Instead, the model can operate on its own, discovering patterns and information that was not previously discovered. It mainly deals with unlabelled data.

Algorithms/techniques:

There are two techniques of Unsupervised Learning

1. Clustering

Clustering is an important concept in unsupervised learning. It mainly deals with finding structures or patterns in unclassified collections of data. The clustering algorithm processes the data and finds natural clusters (groups) if present in the data. You can also modify the number of clusters the algorithm has to identify. You can adjust the granularity of these groups.

2. Association

Connection rules allow you to establish connections between data objects within a large database. This unsupervised technique is to discover interesting relationships between variables in large databases. For example, people buying new homes are most likely to buy new furniture.

Advantages

Unsupervised machine learning looks for all kinds of unknown patterns in your data.
The unsupervised method helps you find features that may be useful for classification.
It happens in real-time, so all input data is analyzed and labeled in front of learners.
It is easier to get unlabeled data from your computer than labeled data that requires manual intervention.

Disadvantages

Exact information about the collation cannot be obtained, and the output is labeled and unknown as data used for unsupervised learning.
The inaccuracy of the results is because the input data is unknown and people haven't labeled them in advance. This means the machine has to do this by itself.
Spectral classes do not always correspond to information classes.
Users should spend time interpreting and labeling classes that follow that classification.
The spectral properties of a class can change over time, so you cannot have the same class information while moving from one image to another.